MAXCC does filtering based on scene meta-data. Scene is 10.000 sq. km, so 10% of cloud coverage might mean that these clouds are fully covering your AOI.

Check this post with ideas on how to do local AOI based cloud filtering:
https://www.sentinel-hub.com/faq/best-way-do-cloud-filtering-specific-aoi

Thank you again for very useful information.

my u.c. is the field of argonomy and i use sentinel S2 for chlorophyl reflection on the fields.

I have a situation where both of the recommended algorithms in the FAQ are not picking up any clouds in an AOI that is almost completely covered by clouds. Is local s2cloudless python processing currently the only way to proceed with this situation or has any work been done integrating s2cloudless into the online dataprocessing scripts for statistical service, etc.?

That sounds strange as the algorithms are not that bad…

s2cloudless is currently not yet integrated in statistical service, so you would have to set-up this python script on your side.

Yeah I was perplexed as well. Those have been working good but this just seems to be an anomaly. s2cloudless picks them up as needed so I can implement that for imagery that doesn’t have the l2a classification like the recent stuff. Thanks!

But what if I my specified bbox placed in multiple Sentinel scenes? Did they calculate average between these two/three/four scenes?

In other words I’m making WMS/WCS request with maxcc=0.3 and bbox placed in two scenes for example. How would the SentinelHub process it?

CC-based filtering happens first (e.g. all scenes in the specified time-range not fitting the criteria are discarded), then PRIORITY kicks in, putting e.g. “most recent” ones on top, then it does the mosaicking.

So, if you have two scenes taken on the same date and one has cc=0.4 and one 0.2, the one with 0.4 will not be shown (or it will be replaced with the older scene, if this is how you set other parameters).